Analytics for AI-Built Mobile Apps: Where to Start

Mobile analytics has more moving parts than web — installs, attribution, store funnels, and platform privacy rules. If your agent just shipped an Expo or React Native app, here's where to start without drowning in SDKs.

Track the lifecycle, not just screens

Screen views are easy and nearly useless on their own. The events that drive decisions map to the user lifecycle:

  • session_started — app opened
  • signup_completed — account created
  • trial_started / subscription_started — monetization funnel
  • purchase_completed — revenue with a value

Instrument those first. You can add granular screen tracking later once the funnel is trustworthy.

Firebase is the sensible default

For most Expo/React Native apps, Firebase Analytics is the pragmatic starting point: free, well-supported, and it feeds the rest of the Google stack. Get the canonical lifecycle events flowing into Firebase before you reach for paid attribution tools.

Attribution is a separate problem

"Where did this install come from?" is answered by an attribution provider (like AppsFlyer), not by Firebase. On iOS, Apple's privacy framework (ATT/SKAdNetwork) limits what you can know — so set expectations: attribution is probabilistic and aggregated, not perfect. Add it when you're actually spending on user acquisition, not before.

Verify on a real device

Emulators lie. Install a dev build on a real phone, walk the funnel, and confirm each event arrives. A purchase_completed that never fires looks exactly like "no revenue" in your dashboard — and on mobile it's easy to miss because the failure is silent.

Keep one taxonomy across platforms

If you also have a web product, use the same event names on both. "signup_completed" should mean the same thing everywhere, so your metrics compare cleanly instead of fragmenting per platform.

Start small: lifecycle events into Firebase, verified on-device, with a shared taxonomy. Attribution and advanced funnels come after that foundation holds.